Photochemical Studies. XLIX. Ketene and Ketene-Oxygen Mixtures1

The Reaction of CH2 (XB1) with O2 (X ): A Theoretical CASSCF/CASPT2 Investigation. B.-Z. Chen ... W. Albert Noyes , Jr. , Gerald B. Porter , and J. Er...
0 downloads 0 Views 888KB Size
Download PDF
VOI. 7(i

3258 [CONTRIBUTION FROM

THE

DEPARTMENT O F CHEMISTRY, UNIVERSITY

OF

ROCHESTER]

Photochemical Studies. XLTX. Ketene and Ketene-Oxygen Mixtures1 BY ALEXANDERN. STRACHAN AND W. ALBERTNOYES,JR. RECEIVED JANUARY 28, 1954 Quantum yields of Zthylene and of carbon monoxide formation have been determined in pure ketene a t three wave lengths: 2654, ?130 and 366c A. Approximately two molecules of carbon monoxide and one of ethylene are formed per quantum a t 2654 A. At 3660 A. the quantum yields are pressure dependent. Thc yield of ethylene is always somewhat less than onehalf of the yield of carbon monoxide. The addition of small amounts of oxygen reduces the ethylene yield to about twothirds of :he oxygen free value a t 2654 and a t 3130 A., but the resultant ethylene yield is independent of oxygen pressurc. At 3660 A. oxygen reduces ethylene formation t o about 26% of the oxygen-free value. Carbon dioxide and formaldehyde together with small amounts of acetaldehyde are also formed when oxygen is present. I t is possible t o suggest two mechanisms which agree with the main facts and to obtain relative rate constants for some of the steps.

The photochemical decomposition of ketene seems to have been investigated first by Norrish, Crone and Saltmarsh.2 These authors showed that carbon monoxide and ethylene are produced in an approximate ratio of 2 : 1 and postulated that ethylene is formed by the reaction of methylene radicals with ketene rather than by association of these radicals. This mechanism was a t one time d i ~ p u t e dbut ~ , ~appears to be confirmed by recent evidence.j+ Further work by Kistiakowsky and his co-workers5 has shown that the ratio of carbon monoxide to ethylene is always somewhat greater than 2 : 1 and that methylene radicals react with ethylene and with carbon monoxide to form propylene and ketene, respectively. They showed, moreover, that quantum yields increase with decreasing wave length. The absorption spectrum of kete?e2p9has a long wave limit between 3700 and 3900 A. Toward the long wave limit there appear to be bands which may be of the predissociation type but which show no evidence of rotational structure. A study of the reactions of methylene radicals with oxygen should have considerable interest both in providing rate constants for someof their reactions and possible clues as to their states. I t became evident as soon as the present work was started that several aspects of the photochemistry of pure ketene needed elucidation before the behavior with oxygen could be understood. Knowledge of the variations of quantum yields with wave length and with pressure seemed to be essential.

Experimental Retene was prepared by pyrolysis of acetone by the method of \Villiams and Hurd.l0 The crude material was distilled at atmospheric pressure in a. Podbielniak low tem(1) This work was supported in part h y contract between the Office of Naval Research, United States Navy, and the Department of Chemistiy, University of Rochester. The authors wish t o express their appreciation t o M r , R. C. Wilkerson, Celanese Corporation of America, Clarkwood, Texas, for performing mass spectrographic analyzes. (2) R . G . W. Norrish, H. G. Crone and 0. D. Saltmarsh, J . Chem. SIJC., 1533 (1933). (3) W. F. Ross and G. B. Kistiakowsky, TEIISJ O U R N A L , 56, 1112 (1931). (4) R. G. W. Surrish, H. G. Crone and 0. D. Saltmarsh, ibid.,66, 1644 (1934). (5) G . B. Kistiakowsky and N. W. Rosenberg, ibid., 72, 321 (1950). (6) M . Vanpee and F. Grard, Ann. Min. Belg., 49, 701 (1950). (7) M. Vanpee and F. Grard, Bull. S O L . ckim. Belt., 60, 208 (1951). '74, (8) G. B. Kistiakowsky and W. L. Marshall, THISJOURNAL, 88 (1952). (9) G. C. Lardy, J . chim. g h y s . , 21, 353 (1924). (10) J. W. Williams and C. D. Hurd, J . Org. Chcm., 5, 122 (1940).

perature distilling column. The middle fraction, which boiled within zk0.lo, was collected and redistilled, the middle fraction being collected and used. The thermocouple was calibrated by distilling propylene (b.p. 47.7'). Ketene was found t o boil a t -50". This boiling point is lower than the previously recorded value (-41') .ll The ketene was stored at the temperature of liquid nitrogen t o prevent slow polymerization. All but 0.0670 was absorbed by acetic acid and of the remainder all but 0.01% was absorbed by potassium hydroxide. Thus the ketene was a t least 99.9% pure. Ethylene was a Matheson Droduct stated t o be 99.5% pure. Non-condensables were removed a t -196". Carbon dioxide (Ohio Chemical Company) was degassed a t -196" and distilled from a bath a t -110'. I t was absorbed completely by potassium hydroxide. Water was distilled from alkaline permanganate, degassed a t - 7S0, and subjected t o low pressure distillation, the middle third being retained. Glacial acetic acid (C.P.) was degassed a t - 78' and vacuum distilled, the middle third being retained. The purification af diethy! ketone has been described.12 Oxygen was prepared by heating C.P. solid potassium permanganate and dried by passage through a trap immersed in liquid nitrogen. Reaction with heated copper indicated better than 9919% purity. Hanovia TvDe SlOO and General Electric Company Type AH-6 lamps were used. A fairly parallel and uni?o& beam was achieved by use of a quartz lens of 10 cm. focal length and a stop. The intensity was varied by neutral density filters.'3 The following color filter combinations were used: (1) Corning glasses 9863 (3 mm.) and 738q (2 mm.). Transmission from about 3400 to about 4300 A. with a maximum (2) Corning glass 5840 (3 mm.) and diphenyla t 3750 butadiene (1.06 mg. in 100 ml. diethyl ether) (4 cm.). Transmission from about 3500 to about 4100 with a maximum 3t 3700 A.; (3) Soda glass, short wave limit abc$ 3100 A.; (4) Pyrex glass, short wave limit about 2800 A.; (5) Corning glass 9863 (3 mm.), NiS04.6Ha0 (500 g./l. in water) (3 cm.) , 2,7-dimethy1-3,6-diazocyclohepta-l,6-dienc iodide (100 g./l.) (2 cm.). Transmission fr2m about 2450 t o about 2960 A. with a maximum a t 2700 A. (6) Corning glass 9863 (3 mm.), chlorine gas (300 mm. pressure) (10 cm.). Transmission from about 2200 to about 2950 A. Thereois also a slight transmission from *about 3700 to about 4200 A. with a maximum about 3960 A. The charactcristics of most of these filters have been described by Rasha.'4 -4 cylindrical quartz cell with plane windows, 20 cm. in length, 2.2 cm. in internal diameter and volume 76 ml. was used. The combined volume of the cell, stirrer and connecting tubing was 218 ml. The light beam filled the cell. Mercury cut-offs were used exclusively and there were no stopcocks in the apparatus. The apparatus was generally of the conventional type used in this Laboratory except for a combined McLeod gage-Toepler pump. The combined McLeod gage and Toepler pump is shown schematically in Fig. 1. An analysis for a mixture of nitro~~

w.;

(11) F. 0. Rice, J. Greenberg, C. E. Waters and R. E. Vollrath, THISJOURNAL, 56, 1760 (1934). (12) K. 0. Kutschke, M. H. J. Wijnen and E. W. R. Steacie, ibid., 74, 714 (1952). (13) See R . Gomer and W. A. Noyes, Jr., ibid., 71, 3391 (1949). (14) .M. Kasha. J . Ogt. SOL.Am., SS, 929 (1948).

June 20, 19.54

KETENEAND KETENE-OXYGEN MIXTURES : PHOTOCHEMISTRY

gen, oxygen and carbon monoxide will be described. The gage i s first evacuated and the-mixture tQbe analyzed allowed to expand into the bulb R. The gas in R is pushed by the mercury into the volume between H a n d a thread of mercury in K. As the mercury rises in R it also rises in U. The float in L is closed as the mercury rises, thus preventing the rise of the mercury into the capillary K. As the mercury is now lowered a thread of mercury is left at H thus confining the gas between H and K. More gas can now be expanded into R and by sufficient repetition of the operation the gas from the line can be brought effectively between H and K. The float in L can now be held down with a magnet and the amount of gas measured as in an ordinary McLeod gage by raising the level of the mercury to D, E, F and G depending on the pressure. The threads of mercury are withdrawn from H and K as the mercury is lowered to the level of the valve in L. The tube containing copper and cupric oxide surrounded by a furnace had been evacuated through Q and T and is now cut off by raising the mercury to S. The trap P is immersed in liquid nitrogen and the cut-off a t N is opened. The valve in L is closed by removing the magnet and the mercury is raised to E. The gas sample is now confined to the space from D through K and the furnace to the expansion bulb Q. I t should be noted that the volume of Q is very large compared to that in the capillary tubing. After several minutes have elapsed to allow for diffusion, the valves in N are closed and the mercury in Q raised to 0. Thirty minutes are usually sufficient for the oxygen to be removed by the copper in the furnace and for the carbon monoxide to be oxidized to carbon dioxide by the cupric oxide. The cut-off is now opened and nitrogen, the only gas remaining, is pumped into the space between H and K with the float in L raised. Its pressure is measured by using R as a McLeod gage. The nitrogen may be discarded through U by removal of the thread of mercury in K. The liquid nitrogen around P is replaced by a Dry Ice-bath and the carbon dioxide formed by combustion of the carbon monoxide is transferred to D and measured. Oxygen is obtained by difference. Other analytical units can, if necessary, be used in place of the copper-copper oxide tube. For each run a small quantity of ketene was remove: from the storage bulb and degassed three times a t -196 . It was allowed to fill the cell to the desired pressure. If oxygen was to be used the ketene was condensed with liquid nitrogen and a measured amount of oxygen introduced. The residual oxygen which did not go into the cell could be measured in the McLeod gage-Toepler pump and subtracted to give the actual amount in the reaction system. A magnetically driven stirrer ensured thorough mixing. I t was found t o be impossible to separate ethylene from ketene a t -183". To effect analyses for ethylene the ketene was a t first allowed to react with water. Glacial acetic acid proved, however, to be preferable. Carbon monoxide and oxygen could be separated satisfactorily from ketene at -196'. A temperature of -215' (supercooled liquid nitrogen) usually was used to ensure that ethylene was not removed simultaneously. After reaction of the ketene with water or acetic acid, ethylene was removed a t -183'. A third fraction volatile a t -160" was also measured. This was mainly carbon dioxide, but for some reason it was found impossible to reduce the blank for this fraction to zero. Analyses for carbon dioxide are, therefore, less accurate than those for oxygen, carbon monoxide and ethylene but should be good to about twenty %. Analytical procedures were checked by adding to ketene known mixtures of oxygen, ethylene and carbon dioxide. Quantum yield determinations were made in two differect wave length regions, uiz., about 2700 and about 3700 A. For the determinations a t 2700 A. diethyl ketone was used as an internal actinometer. The yield of carbon monoxjde per quantum was assumed t o be unity as it is at 3130 A." The variations of the light intensity during a run and from one run to the next were measured by a Type 935 General Electric Company phototube and galvanometer. The photocell-galvanometer combination was used likewise to obtain relative quantum yields a t 3700 A. The spectral response curve of the Type 935 phototube supplied by the General Electric Company waso used t o convert relative yields to absolute yields a t 3700 A. since no internal (15) 1 %'.

Davis, J r

, THISJ O U R N A L ) 70, 1868 (1048).

To Analytical Line

3259

To Cell

Line a i d A

Fig. 1.-Diagram

Vacuum

of the combined McLeod gage-Toepler pump.

actinometer is available a t that wave length. From the curve the sensitivity (in tefms of equal radiant flux) a t 2700 A. is 78% and a t 3700 A. is 95% of the maximum sensitivity at 3200 A. For equal numbers of quanta the sensitivity a t 3700 A. will be (95 X 2700)/(78 X 3700) 0.89 times the sepitivity a t 2700 A. From the actinometer runs a t 2700 A. and this sensitivity*factor the absolute magnitudes of quantum yields a t 3700 A. could be obtained.

-

Results There was neither measurable thermal decompoposition of ketene nor thermal reaction with oxygen a t any of the temperatures and pressures used in the present work. There is, apparently, a very slow thermal dimerization or polymerization but the pressure change during blank runs amounted to less than 1% of the total pressure. The intensity of the SlOO Alpine burner was very constant although there was a slow decrease with time, amounting to less than 1% per hour. By interspersing runs with and without oxygen i t was possible by interpolation and extrapolation t o obtain ratios of rates of product formation with oxygen present to those with no oxygen. The AH-6 lamp was subject to large fluctuations in intensity. The intensity was measured a t frequent intervals with the photocell-galvanometer combination and the mean intensity during a run obtained graphically. Calculations of ratios of rates with oxygen to those without are corrected for

TABLE I PHOTOLYSIS OF KETESE-OXYGEN MIXTURES ( R (in molecules X l01*/sec.) is the number of molecules of CO formed per second in pure ketene under the same cotiditions of temperature, pressure and intensity) as for deterniination of the other rates which are expressed in the same units.) Lamp-filter comb. and wave 1.

Time ,frun, mm.

T 3

:I Li

I

1

b 1

a

e" (2700) d (2700) d (2700)

60 30 30 00 60 300 60 300 60 120 120 120

=

R

Initial and final 02 press. m m . X 103

Rco/R

RC~E,/R

26 =t2O, P (ketene) = 39 mm. unless otherwise specified 1.33 54(12) 1.29 0.33 57(9 6 ) 1.14 .. 2.07 2.10 100(65) 1.30 .. 1.35 158(108) 1.25 .31 1.42 484( 440) 1.30 .33 1.29 491(267) 1.30 .. 6 . 15 1037(811) 1.32 .35 1.10 2127( 1881) 1.41 .33 1.40 2278(2220) 1.28 .31 1 .$7 540( 460) 1.08 .32 2.96 2096( 1889) 1.30 .37 2.40 20 mm.(20 mni.) 1.33 .32

0.62 .60

.BO .73 .50 .F7 .72 .83 .82 .48 .09

..

.. ,23 .34

.. .19 .19 .45

T = 107 f l o ,P (ketene) = 39 mm. unless otherwise specified 90 8.23 991(251) 1.05 .48 1.39 .30 45 8.53 1198(677) .61 1.42 1.59 .33 b 75 5.58 1156(614) 1.36 .. 1.64 .. b 90 0.76 1021(925) 1.47 .07 1.48 .32 b 15 5.45 1058(1)68) 1.52 .. 1.59 .. a 60 1.20 1009(888) 1.75 .. 1.51 .2s a 00 1.16 1.58 .80 1009(904) 1.52 .2s C 180 0.27 3.00 1.19 1044(903) 1.83 .I7 r (3700) 30 2.71 1.14 2.81 1042(821) 2.02 .05 fd (3700) 30 1.43 4.45 2.23 1027(845) 2.54 .11 b 10 5.89 10.7(52) .. 0.95 1.44 .36 1, 60 6.08 1.06 .. 478(108) 1.53 .32 b 60 5.94 1.36 .. 1811(1349) 1.60 .31 Note: The ratios in columns 5-8 are obtained by comparison with blank runs for which the wave length distribution was the same as for the oxygen runs. a P (ketene) = 94 mm. * P (ketene) = 92 mm. P (ketene) = 143 mm. P (ketene) = 140 mm LVave lengths are given only in case the radintion was approximately monochromatic. d (2700) dC (2700)

variations in intensity, but ratios with the AH-6 lamp are subject to greater errors than with the SlOO and should not be relied upon to better than ten or possibly even 20%. Oxygen was not added during runs and hence its pressure decreased continuously. It transpired that any error due to this source is relatively unimportant because of the lack of variation of rates with oxygen pressure. Formaldehyde and acetaldehyde were determined by the method of Stotz16as modified by Erbelding and Walters. l7 The following symbols are used in the Tables for various lamp-filter combinations : (a) S l O O l a p p , filter 4 (all wave lengths greater than about 2850 A.) ; (b) S l O O lamp, no filter (all wave lengths greater than about 2400 k . ) ; (c) SlOO lamp, filter 3 (all wave lengths greater than about 3200 k.); (d) AH-6 lamp, filter 6 (wave lengths around 2700 k . ) ; (e) iZH-G lamp, filter 5 (wave lengths around 2700 A.) ; (f) AH-6 lamp, filter 1 (wave lengths around 3700 k ) . Five runs were made a t 26 =t2" with lamp-filter combinations a, b, d and e at pressures of 39 and 143 mm. without oxygen. Rates of CO formation varied from 1.43 to 6.22 X molecules/sec. The (16) E. Stotz, J . B i d . Ckem.,148, 585 (1943). (17) W. F. Erbelding and W. D. Walters, unpublished results, University of Rochester. See M. Christie, THISJ O U R N A L , 76, 197!) ll05.li

ratio of ethylene to carbon monoxide was 0.4GS f O.OOG with no trends apparent. A t 107 f 1" nine runs were made without oxygen and with lamp-filter combinations a, b, c, d and f and pressures of 39 and about 140 mm. The value of Rc~H~,/Rco was found to be 0.453 f 0.012 with no trends apparent. This ratio is, therefore, essentially independent of temperature, pressure, wave length and rate of carbon monoxide formation. No propylene was found among the products. Table I presents the data for ketene-oxygen mixtures. It can be seen from the data in Table I that if the ketene-oxygen reaction is {ue primarily to radiation of wave length 3130 A. or shorter the ratio Rc,HJR is0.33 =k 0.01 a t 26' and 0.31 i 0.02 a t 107'. This ratio is, therefore, independent of oxygen pressure, of ketene pressure, and of temperature within experimental error. The effect of oxygen is to reduce the rate of ethylene formation to about 70% of the oxygen-free value provided short wave lengths are the main ones causing reaction. I t is evident also from Table I that if the radiation causing reactipn is mainly of wave length longer than 3130 A. (lamp-filter combinations c and f) the ratio Rc%H~/R is low, ;.e., about 0.1 and the rate of ethylene formation is reduced to about 25% of the oxygen-free value. These facts seemed to indicate a fundanientnlly different behavicir at

KETENEAND KETENE-OXYGEN MIXTURES : PHOTOCHEMISTRY

June 20, 1954

3700 A. from that at 3130 A. and shorter wave lengths. It seemed desirable to make quantum yield determingtions in oxygen-free ketene a t 2700 and a t 3700 A. Table I1 shows data obtained a t the former wave length a t two temperatures, 26 and 107", andoFig.2 shows plots of ~ / @ . cos. o P (ketene) for 3700 A. a t four temperatures: 26, 27, 107, 154'. The data shown in Fig. 2 were obtained over a range of intensities so that the number of quanta absorbed per second was a minimum of 6.3 X 1014 and a maximum of 156 X There was no significant trend in 9.~0 with intensity. QUANTUM YIELDS Ketene pressure, mm.

OF

TABLE I1 co FORMATION IX P U R E KETENEAT 2700 A.

Quanta absorbed X lO-"/sec.

Rco

326 1

P

7

/

150

26T.

. 100 8 2 9

I

molecules X lO-"/sec.

Qco

T = 26' 39 143 143

0.392 0.780 1.60

39 143

0.398 r. 11

0.756 1.76 3.45

1.93 2.26 2.16

60

T = 107' 0.770 2.m Average of all

1.93

2.33 2.12 f 0.15

The quantum yield of carbon monoxide formation in pure ketene is 2.0 within experimental error a t 2700 A., and there is probably no significant variation with temperature. There may be a trend with pressure but it scarcely exceeds experimental error. These results arc in marked contrast with quantum yields a t 3700 A. (Fig. 1). The latter are very much lower and they decrease a t high pressures and increase a t high temperatures. These facts show, as do the data in Table I , that there is a marked difference in behavior between short wave lengths and long. The data in Table I show that oxygen causes an increase in carbon monoxide formation and a decrease in ethylene formation a t all wave lengths and temperatures. The rates of oxygen disappearance are such as not to indicate a long chain reaction under any of the conditions used. There is some tendency for the rate of carbon dioxide formation to increase with increase in oxygen pressure, but in no case is this tendency very marked. For the fifth run a t 107" (Table I), RHCHO/R was found to be 0.24 and R C H ~ C K O to/ be R 0.04. The radiation during this run was polychromatic but that which caused reaction was mainly a t 3130 A. and shorter wave lengths. Under these conditions acetaldehyde is formed in small quantities and the quantum yield of formaldehyde is roughly 0.05. The conclusions which may be drawn from the mass spectrographic analyses are as follows: (1) More acetic anhydride is formed when oxygen is present than when i t is not. This indicates that water is one of the pmrliirfs of the reaction, ( 2 ) Formaldehyde and acetaldehyde are formed in small amounts. ( 3 ) No formic acid is found. The quantitative determination of formaldehyde and of acetaldehyde would not have been possible by the mass spectrograph due to polymerization and to

I

,

I

4.0

1.0

(Ketene) molecules/ml. X 10-l8. Fig. 2.-Plot of the inverse carbon monoxide quantum yield us. concentration a t 3700 A. and several temperatures.

the necessary conditions for collection and shipping of samples.

Discussion It has been indicated that there are fundamental differences in behavior of either pure ketene or of mixtures of ketene and oxygen between those wave lengths which lie near the long wave absorption limit and those which are much shorter. For this reason it is logical to divide the discussion into three parts: (a) pure ketene a t short wave lengths; (b) ketene-oxygen mixtures a t short wave lengths; (c) pure ketene and ketene-oxygen mixtures a t long wave lengths. During the early stages of this work the importance of wave length was not fully appreciated and polychromatic radiation was generally used. By comparison of these results with later results obtained with approximately monochromatic radiation it is possible to decide which radiation was mainly responsible for the reaction. Thus these early results have some value in aiding- the interpretation. (a) Pure Ketene at Short Wave Lengths.Positive proof concerning the nature of thephotochemical primary process does not exist, but the accumulation of evidence from all sources,2-D including the results of tellurium mirror removal of Pearson, Purcell and Saigh,l8 indicates very strongly that it may be written CHXO

+ hu = CHI + CO

(1)

(18) T G Pearson, R H Pl~rcelland G S Saigh. J Chcm. S O L , 409 (193SJ.

ALEX~INDER S . STRICHAX ASD UT.;ILDERT SOUES, JR.

3262

1.01. 76

exactly unity, the quantum yield of carbon monoxide must be exactly two, a value which would disagree with the CO, C2H4ratio of 2.2 found expcrimentally. The nearer @C.II~is to unity, the higher the average value of n must be in the forinula of the postulated products (CH?),,CO. The data in Table I1 are not accurate enough to permit more detailed CII2 -L CH2CO = C J I i + CO (2) conclusions to be drawn. The results in Table I1 show the quantum yield of There niay be a relationship between one oi the carbo? nionoxide formation to be about two a t two forms of (CH2)2C0(possibly CHrCOCH?)and 2700 A . This is the value to be expected if reac- allene formation found by Kistialwwsky aiicl tions 1 and 2 were the sole oqes. Thus the priFurther work on this point might yield mary quantum yield a t 2700 A. is probably very interesting results. (The authors are indebted to close to unity, a value not clearly indicated by Professor Kistiakowsky for this suggestion.) earlier work. (b) Ketene-Oxygen Mixtures at Short Wave The reaction CH2 (CH2)2CO = CzH4 Lengths.-The data in Table I show clearly that CH-CO postulated by Kistiakowsky and co-work- whenevcr oxygen is present and the rextion is due crsj8 is not necessary if the quantum yield of car- predominantly to wave lengths of 3130 A. or bon monoxide formation is two. Reaction 2 is shorter, the ethylene yield is reduced to about 0.22 sufficient to explain ethylene formation. On the of the oxygen-free value and that thc yield does not other hand, it is necessary to postulate the radical vary either with oxygen pressure, if the oxygen (CH7'lrC0t o explain the formation of compounds of pressure exceeds about 0.01 mm., or with teinperahigher molecular weight (see below). ture. The carbon nionoxide yield is increased Less than 1% of the ketene was allowed to de- about 30mo by addition of oxygen a t 26" and about compose in each of the experiments herein 6Op0 a t 107". At both temperatures there i i i ~ ybe described. Under these conditions reactions of a slight tendency for the CO yield to increase with CH, radicals either with ethylene5 or with carbon oxygen pressure, but the trend is hardly more than monoxides may be neglected. This is confirmed by experimental error. the absence of propylene in the products. In The only measured product whose yield seeills to spite of these facts the average ratio of carbon mon- increase with oxygen pressure is carbon dioxide, and oxide to ethylene in the products is 2.20, in excellent its yield also increases with increase in temperature. agreement with 2.19 found by Kistiakowsky and The formation of large amounts of ethylene in the his c o - w ~ r k e r s . ~ presence of oxygen is the surprising fact in these The ratio must be due to reactions in addition to results. 1 and 2 and whose rates are essentially independent The following numbers of molecules are fonned both of ketene prcssure arid of temperature. They or disappear per methylene radical produced in niay be pure ketene: 2F", CO, 2.6; C2H4, 0.64; 0 2 conCH? + CHaCO = (CH2)zCO ( 31 sumed, 1.4; COz, 0.4; 107", CO, 3.2; CzH4, 0.64; (CH*)?CO CHzCO CO (CH?)aCO ( 4 ) O2 consumed, 2.8; COz, 1.2; HCHO, 0.48. These yields are alniost constant as long as the ratio Continuation of reactions of the type of 4 could lead to compounds of higher molecular weight with CHzCO 'Or does not exceed about ZOO. These results seem to make inescapable OIIC or the empirical formula (CH2)llCO (crotonaldehyde the other of the following conclusions. (1) methyland, or cyclic ketones may be formed). The radical formed by 3 may be either CH.CH2- ene radicals react so rapidly with ketene that their CO or CH2COCH?. Possibly one of the two is re- reaction with oxygen is unimportant a t the pressponsible for ethylene formation either by direct sures used; (2) methylene radicals react with oxygen so rapidly that their reaction with ketene is undissociation or by reactions of the type important even a t the lowest oxygen pressures (CHs)2C0 + C&CO = C2H4 + CO + C H C O (3) used. If 1 is accepted, it must be assumed that and the other for coinpounds of higher molecular the radical (CH2)*C0reacts with oxygen about oncthird of the time and reacts to give ethylene about weight by the over-all reaction (CH2)zCO (72 two-thirds of the time (presumably without con2)CH*CO = (CH2)nCO (a - 2)CO. If E = 3, the ratio of the rate of the initial step leading to sumption of oxygen). On the assumption t h a t all hydrogal- atoms form ethylene to that of the other step would be 10: 1. Ai water a fairly good material balance is obtained high value of n means a high value of the ratio. Positive identification of compounds formed in both a t 26" and a t 107", although slightly more oxygen is consumed than can be accounted for in the small amounts would be desirable. Two possibilities must be considered: (1) only products. It is necessary, however, to assume that one intermediate is formed, and it may react either slightly more than two molecules of ketene disapto give ethylene or to give a polymer; (2) two dif- pear per quantum absorbed a t 26" and slightly more ferent intermediates are formed and react, respee- than three a t 107". This suggests that me are tively, with ketene to give ethylene and a compound dealing with a short chain reaction with a chain length which increases with temperature. It is of higher molecular weight. It should be pointed out that 'iccording to the possible to write a logical sequence of steps based mechanism proposed the quantum yield of ethylene on the assumption that CH2 radicals do not react formation may not exceed unity. Indeed, if it is directly with oxygen, that (C112)2C0r,idic,ils o f The magnitude of the primary quantum yield is evidently dependent on wave length and a t the longer wave lengths is also dependent on pressure (see Table I1 and Fig. 2 ) . Previous work"? indicates that the principal secondary reaction may be written

+

+

+

+

+

+

June 20, 1954

KETENEAND KETENE-OXYGEN MIXTURES : PHOTOCHEMISTRY

one type always give ethylene and that (CH&CO radicals of the other type react with oxygen to give the other observed products, with additional ketene molecules being involved in the chain. These steps would be purely speculative, and there is no need to pursue the subject further a t the present time. If assumption ( 2 ) is accepted, however, so that CH. radicals always react with oxygen, i t is necessary to assume further that ethylene is formed by CHz

CH202

+

0 2

= CHzO?

+ CHzCO = CzH4 + + CO 0 2

(6)

(7)

Reaction between CH20iand ketene must also lead to other products, possibly depending on the nature of the intermediate complex. Kistiakowsky and Rosenberg5have estimated the rate constant of reaction (2) to be of the order of magnitude of ml./molecule sec. Tne c a l d a tion of this constant was based on the assumption that mirror removal is brought about solely by CH2 radicals. If diradicals such as (CH&CO may also remove mirrors, ml./molecule sec. would be a lower limit for the constant of reaction ( 2 ) . Reaction (6) must have a rate constant a t least 500 times greater than that of reaction ( 2 ) if assumption 2 is adopted, ;.e., it must have a value greater than 5 X If (6) has zero activation energy, the effective cross section would be about 3 x cm.2. If the collision diameter is taken as 4 A,, the steric factor would be of the order of magnitude of lo+. The value of 2 X was found for the reaction of methyl radicals with oxygen.lg Thus there seem to be two ways of explaining the reactions in the ketene-oxygen system. One possibly valid objection to assumption 1 can be made. In the absence of oxygen it is necessary to assume that approximately 10% of the methylene radicals react with ketene to give the non-ethylene forming radical, whereas in the oxygen system it would be necessary to assume that approximately one-third of them do this. Further work may distinguish between the two mechanisms provided satisfactory experiments can be devised for the purpose. It is interesting to note that ethylene is also formed in the presence of nitric oxide.20 (c) Ppre Ketene and Ketene-Oxygen Mixtures at 3700 A.-The very marked effect of pressure on the quantum yields in pure ketene is shown in Fig. 2 . The equations of the straight lines are all of the form l/@ = a bP. Possibly several methods of deriving an expression of this type could be visualized but one is much more plausible than the others

+

CHzCO

CHzCO

+ CH2CO'

CHzCO' CHzCO'

+ hv

= CHnCO'

(8) 2CHzCO (9) = CHz CO (10) = CHZCO (+ h ~ ? (11) )

+

followed by reactions of the methylene radicals already discussed. Since the CO/CZH~ratio is independent of pressure, temperature, intensity and (19) F. B. Marcotte and W. A. Noyes, Jr., Discussions Faraday Soc., No. 10, 236 (1951). (20) M. Burton, T.W. Davis, A . Gordon and H. A . Taylor, THIS JOURNAL, 63, 1956 (1941).

3263

wave length, it may logically be assumed that methylene radicals pnce formed will react in the same way a t 3700 A. as a t shorter wave lengths. Thus by assumption of the steady state one finds Ia/Rco = l/*ro = (1/2)(1

+

(h/klo)

f ki(CHCO)/kio) (12)

where I , = number of quanta absorbed/ml. sec., Rco = rate of formation of CO in molecules/ml. sec., 2 = number of CO molecules produced per methylene radical formed in the primary process a t 2700 @CO = quantum yield of carbon monoxide formation. Equation 12 is of the right form to fit the straight lines in Fig. 2. (NOTE: Even higher pressures than those shown in Fig. 2 were investigated a t 26". These points fall on the same linear plot.) Reaction 11 must be included bgcause of the very low quantum yields a t 3700 A. These are much lower than those reported previously for this wave length.3 The intercepts found from equation 12 are kll/k,o) and the slopes (k9/2 k ~ o ) . It is seen ( tnat both the slope and the intercept increase with temperature. If the rate constants are all written in the form k = a exp(-E/RT)

a.,

+

k9/2k10 = ( a d 2 a 1 0 exp ) ( - ( E $ - Eio)/XT)) (13)

abd it is possible to calculate E,, - El" from a plot 0; the logarithm of the slope us. 112". The result is ( E 9 - E&,) = -4500 =t300 cal. per mole. Simi2) X ml /molecule. If larly (aglalo) = (4 reaction 9 has a steric factor of unity, nlo is about 1O1O sec.-l. This is an upper limit. If E9 = 0, El0 = 4500 cal. per mole would be the activation energy for dissociation of active molecules. This is of the same order of magnitude as that estimated for a similar step in acetone.*I The data necessitate two unimolecular methods of disappearance of CHLCO', uiz., (10) and (11). Possibly (11) results in fluorescence but if so the emitted light has a wave length so close to that of the absorbed radiation that it could not be detected. Further work on this point would be desirable. In any case the intercepts provide a method of determining klo/(klo All). X plot of In (2i-I) vs. 1/T should have a slope proportional to El0 - E11 (i = intercept in Fig. 2). The intercepts cannot be determined with sufficient accuracy to permit good determination of this activation energy difference but it proves to be slightly less than El0 - Eg. There may be, therefore, a slight activation energy (perhaps 1000 cal.) for reac$ion 11. Two runs were made a t 107" a t 3700 A. with added oxygen. Oxygen a t this wave length markedly reduces the yield of ethylene, carbon monoxide is greatly increased, carbon dioxide is formed, and the oxygen consumption is relatively large. These facts indicate a differentoreaction a t this wave length from that a t 2700 A. The obvious conclusion is that the active molecules react with oxygen

*

+

CHzCO'

+ 02 = (no ethylene)

(14)

The small amount of ethylene-formed in the presence of oxygen could be due either to some stray (21)

W A Noyes, Jr

788 (1948).

and L I\

Dorfman, J Chern P h y s , 16,

C. REID

3264

radiation of shorter wave lengths or to competition between (10) and (14). If it is assumed that there is no stray radiation an estimate can be made of k14/k9. This gives 323 and 316 for the two runs. Thus if every collision of an activated molecule with oxygen is effective in preventing formation of methylene radicals, the steric factor for the deactivation by ketene must be not greater than 0.003. This would make a10 = 3 X lo7sec.-l, rather than 1O'O as given above. A second method may be used to estimate k14/k9. From the data in Table I1 it is found that 2.8 oxygen molecules are consumed for each ethylene molecule formed a t 2700 A. The remaining oxygen must be consumed by reaction 14 and the reactions which follow it. If it is assumed that only one oxygen molecule is used as a result of (14) one can obtain klJklo and combine this with ks/kro to obtain k14/k9. The two runs give 77 and 54, respectively. Of the two methods the second seems to be based on the more doubtful assumptions. One may take a10 to be not more than lo8 sec.-l so that klo < IO8 exp(-4500/RT). At 107' Kl1/ klo -8 (Fig. 2) and EIO- Ell 3500 cal./mole. Hence all < lo7 set.-' and the mean life will be greater than sec. A calculation of the mean life based on the integrated absorption coefficient gives 3 X sec.22 The approximate agreement

-

(22) We are indcbted t o hlr. H K. Dice and Dr. 0. V. Luke of the

[CONTKIBU'IIOH PKOM

VOl. 76

between the mean lifetimes obtained by the two methods may be fortuitous but does suggest the desirability of a more careful investigation of the spectrum and possible fluorescence of ketene vapor. These mean lifetimes would be compatible with existence of rotational structure in the ketene spectrum. More detailed work will be necessary to establish the nature of the reaction of excited ketene molecules with oxygen but i t may be said with considerable certainty that such a reaction does occur. The results presented in this paper do not establish conclusively the detailed mechanism of the reactions of methylene radicals with oxygen largely because the lack of variation of the results with experimental conditions prevents use of the customary methods of competitive reactions. It may be concluded that in the case of ketene, and probably also for other molecules, i t is necessary to exercise considerable care in experimental procedure and in interpretation to be sure that reactions of active molecules with oxygen are not confused with reactions of radicals. Celanese Corporation of America, Clarkwood, Texas, for permission t o use absorption coefficients obtained in their research laboratory b y Messrs J. S. Hill a n d R. C. Wilkerson. T h e calculation of t h e mean life was made by Mr. Richard Holroyd a t t h e suggestion of Professor A. B. F. Duncan.

ROCHESTER, XEWYORK

LAI3ORATURY O F MOLECULAR STRUCTURE A N D SPECTRA, DEPARTMENT O F rrIIE UNIVERSITY O F CIIICACO]

1'HYSICS,

The Aromatic Carbonium Ions' BY C. RE ID^ RECEIVED FEBRUARY 1, 1954 The spectra of a number of aromatic hydrocarbons dissolved in liquid hydrogen fluoride containing boron trifluoride have been examined. Two kinds of absorption bands have been observed arid are attributed (1) t o the aromatic carbonium ions R H + and (2) to the complexes R : BFI. The carbonium ions are subdivided into two groups with spectra centered a t about 4,000 and 4,800 A,, respectively, and a tentative explanation for these is put forward. Photochemical changes have also been observed on irradiation of some of the polycyclic carbonium ions and are discussed.

Introduction The fact that the more basic, polynuclear hydrocarbons dissolve in strong sulfuric acid, and that even the less basic ones, benzene, toluene, etc., will dissolve in anhydrous hydrofluoric acid, in the presence of boron trifluoride, is ~ e l l - k n o w n . ~Differ~-~ ences in basicity have been used as a mode of separation of these hydrocarbons.3 Nevertheless, data on the "carbonium ions" which are formed, it is thought by the addition of a proton to the hydrocarbon-which thus acts as a base-are very scanty, although some spectra of sulfuric acid solutions of the more basic hydrocar(1) This work was assisted by t h e Office of Ordnance Research under Project TR2-0001 (50.5) of Cnritrort D.~.11-022-ORD-I00? w i t h the University of Chicagn. ( 2 ) On leave of absence from t h e University of British Colnmhirr, Vancouver, B. C., Canada, which is presently t h e correct address. (3) (a) M. Kilpatrick and F. E. Luborsky, THISJ O U R N A L , 75, 577 (19n3). a n d references given there: ( h ) n. A . M c C a u l a v , R . II. Shoeni.ikrr , i i i i l .\ I' 1 icn. / : ! ! l : ! ~ (: ' i . i . i i , 4 2 , 2103 ( I ~ W ~

bons have been p ~ b l i s h e d . ~The experimental work here described shows that in fact the phenoinena occurring in HF-BFa solutions are quite involved, addition compounds of BF3 and hydrocarbon sometimes forming as well as the carbonium ion, which may itself react to form further products. Experimental Absorption.-Anhydrous H F (Mathieson C.P.) and dry boron trifluoride (from the same source) were distilled through Teflon tubes directly into the absorption cell which was kept a t -80' during the distillation and subsequently. I n some of the earlier runs the HF was distilled from a small copper cell and the middle fraction only used. However, no difference in transmission could be detected between this material and that distilled directly from the supply cylinder. It showed no trace of the absorption band a t 2800 characterijtic of SOZ, and was assumed substantially free of that material. BFI was distilled into the cell through a cold trap kept a t -100" by a pentane-bath. It was assumed ( 4 ) V. Gold and (liircs therrin.

F. L. T y e , J . Chcm. S o c . , 2172 (1952). and refer-